Collaborative Proposal: Frameworks: Project Tapis: Next Generation Software for Distributed Research

强大的网络基础设施（CI）生态系统的目标是通过促进软件框架作为可持续生产质量服务的发展，成为发现和创新的催化剂。现代科学和工程研究越来越多地跨越多个地理分布的数据中心，并利用仪器、实验设施和国家和地区CI网络。Tapis框架将使科学家能够以一种安全、可扩展和可重复的方式完成计算和数据密集型研究，使科学家能够专注于他们的研究，而不是完成研究所需的技术。Tapis将允许更容易地实现、共享和重用复杂的计算应用程序、工作流和基础设施，并使以前对研究人员来说过于具有挑战性的分析成为可能。该框架将最大限度地提高应用程序的可移植性，允许灵活地调度地理分布的计算工作负载，提供基于web的科学即服务，以实现多设施、分散部署，并为传感器和流数据提供生产级支持。Tapis将影响多个科学领域、地理和代表性不足的社区，这些社区有可能解决世界上最重要的科学问题，包括天文学、气候科学、医学、自然灾害和可持续性科学。教育和推广将包括赞助研讨会、黑客马拉松和涵盖平台的培训材料，并提供示例，以鼓励用户广泛采用各种技术技能，并通过沉浸式研讨会和专业发展机会瞄准下一代年轻研究人员和专业人士。Tapis将是一个分布式计算实验的新平台，利用NSF在Agave、Abaco和CHORDS项目上的投资。Tapis软件框架将1)为传感器和流数据提供生产级支持，2)最大化应用程序可移植性，允许跨地理分布提供商灵活调度计算工作负载，3)提供基于http的科学即服务RESTful api，以实现多设施、分散部署，既安全又可扩展。Tapis将与不同领域的研究人员一起推动现实世界的用例，成为计算工作流和科学网关的底层网络基础设施。Tapis将利用容器来最大限度地提高应用程序的可移植性，允许跨地理分布提供商灵活地调度计算工作负载。项目将通过在整个框架中引入执行系统功能和应用程序需求来实现这种灵活性。Jobs服务将以分布式方式运行，以利用数据局部性，并可选地在未充分利用的系统上调度作业。Tapis将使用微服务架构部署在集中式或分布式配置中，该架构包括一个新颖的、分散的安全和授权内核。该内核可以部署在本地，以保留对机密数据的本地控制。自定义微服务可以插入到安全内核中以提供新的功能，从而形成分布式计算的网络基础设施生态系统。为了有效地执行Tapis，来自德克萨斯高级计算中心（TACC）、德克萨斯大学奥斯汀分校（UT）和夏威夷大学（UH）的团队将利用长期合作来支持研究者驱动的、地理分布的、数据密集型的研究。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The goal of a robust cyberinfrastructure (CI) ecosystem is to become a catalyst for discovery and innovation by fostering the development of software frameworks as sustainable production-quality services. Modern science and engineering research increasingly span multiple, geographically distributed data centers and leverage instruments, experimental facilities and a network of national and regional CI. The Tapis framework will enable scientists to accomplish computational and data intensive research in a secure, scalable, and reproducible way allowing scientists to focus on their research instead of the technology needed to accomplish it. Tapis will allow easier implementation, sharing and re-use of complex computational applications, workflows, and infrastructure and enable analysis previously too challenging for researchers. The framework will maximize application portability, allowing flexible scheduling of geographically distributed computational workloads, offer a web-based science-as-a-service to enable multi-facility, decentralized deployments, and provide production-grade support for sensors and streaming data. Tapis will impact multiple science domains, geographic and underrepresented communities with the potential to tackle the world's most important scientific problems spanning astronomy, climate science, medicine, natural hazards, and sustainability science. Education and outreach will include sponsored workshops, hackathons and training materials covering the platform and providing examples to encourage widespread adoption for users across a variety of technical skills and targeting the next generation of young researchers and professionals through immersive workshops and professional development opportunities.Tapis, will be a new platform for distributed computational experiments that leverages NSF's investments in the Agave, Abaco and CHORDS projects. The Tapis software framework will 1) provide production-grade support for sensors and streaming data, 2) maximize application portability, allowing flexible scheduling of computational workloads across geographically distributed providers, and 3) provide science-as-a-service HTTP-based RESTful APIs to enable multi-facility, decentralized deployments that are both secure and scalable. Working alongside a diverse set of domain researchers to drive real-world use cases, Tapis will be the underlying cyberinfrastructure for computational workflows and science gateways. Tapis will leverage containers to maximize application portability, allowing flexible scheduling of computational workloads across geographically distributed providers. The project will achieve this flexibility by introducing execution system capabilities and application requirements throughout the framework. The Jobs service will be run in a distributed manner to take advantage of data locality and, optionally, to schedule jobs on underutilized systems. Tapis will deploy in centralized or distributed configurations using a microservices architecture that includes a novel, decentralized security and authorization kernel. This kernel can be deployed on-premises to retain local control over confidential data. Custom microservices can be plugged into the security kernel to provide new capabilities, resulting in a cyberinfrastructure ecosystem for distributed computing. To effectively execute Tapis, teams from the Texas Advanced Computing Center (TACC), the University of Texas at Austin (UT), and the University of Hawaii (UH) will leverage a long-standing collaboration to support investigator-driven, geographically distributed, data-intensive research.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.